Nice article, and although almost five years old and describing a CPU and chipset architecture that is no longer used, the Vcore, Vdroop, VID, and LLC concepts are basically the same today.

It's interesting how basic knowledge like this is not part of what is usually included in Intel CPU over-clocking tutorials. The main reason for that IMO is based on the generally accepted and deeply ingrained concept that what Intel tells us is true, versus what has been found to be true, are two different things. In other words, the rules set by Intel can be and are broken all the time, without any consequences. That is true in many cases, but not in all. The ability to get away with things has made users and mother board manufactures bolder in what they do, but there are limits and we have been more so lucky than smart.

For example, the memory speeds used with SB and IB CPUs are beyond the Intel specs, and openly marketed by board and memory manufactures.

Another example, over clocking via the Turbo multiplier is not meant to be constant, but time limited based on several conditions. Settings to break this rule are well known to board manufactures, and provided in the UEFI/BIOS as default settings. We can run our CPUs at 4.5GHz+, with memory operating at 2133 or beyond, and all is well as long as we cool the CPU enough.

I also don't see a droop in Vcore under load, apparently, but I use the CPU power saving options, so my Vcore is all over place, and not set to a specific value.

The explanation of Vdroop in the article makes sense, and the potential dangers of LLC clear. Or has VRM technology become fast enough to prevent voltage overshoot? Probably not...